There’s a well-known saying in finance: “Past performance is not indicative of future results.”

Though that’s an oft-ignored disclaimer, it’s also wise advice when it comes to a lot of things, including online media. As someone who is immersed in that world, I can attest that hot websites come and go and it’s not unusual for a new site to come out of nowhere and quickly dominate views and discussion on a given topic.

We’ve all seen this phenomenon in social media. In 2006, MySpace was so big that it had more visitors in the U.S. than Google. These days, the network gets 50 million visitors a month, which sounds impressive until you compare it with Facebook’s 1.4 billion.

More recently, a bunch of sites including the BBC, BuzzFeed, The New York Times and The Huffington Post,saw a huge dropoff in unique monthly readers in April, though they all recovered later on.

Since traffic fluctuates so often, in this media environment it’s important to take an improvisational, experimental attitude that combines machine learning and human intuition. You need an algorithm that can identify hits, but you also have human interlocutors to judge a site’s momentum. A machine can’t tell when a media property has jumped the shark, but human beings can.

The Human Element

For example, a machine won’t note or anticipate the impact of a new editorial hire. The truth is, though, media properties don’t run themselves but are instead the result of human effort.

One of the world’s top sites is BuzzFeed, but that’s due in a large part to the backing of Jonah Peretti, one of the founders of The Huffington Post, who started BuzzFeed in 2006 as an experiment of sorts. In recent years, as BuzzFeed has grown, it’s attracted some top industry talent, including Ben Smith of Politico and, more recently, John Paczkowski, formerly of Re/Code, whose consistent Apple scoops have made the site a must-read for tech news. In digital, momentum shifts at warp speed. When you starting landing the big stories and connecting with the Zeitgeist, people notice, fast. The reverse is also true.

Time Inc., for instance, was riding high in the mid-aughts with InStyle, Real Simple and CNN Money. In fact, the company was often used as an example of an old school company that was able to deftly transition to the digital world.

Who’s Next?

Given the oft-ignored back stories, media buyers should again take a cue from the world of finance. Companies aren’t self-sustaining organisms but groups of people aligned around a common goal. The hiring of a new CEO can make a huge difference, just ask Apple. But new talent is constantly emerging as well. That’s why it’s important to cast a wide net for new, emerging hits. It’s also a good idea to take a close look at inventory which has performed well in the past to see if it’s still measuring up.

This is a process that you can’t automate. When placing media buys, it is essential to be nimble, adjusting for changes in traffic and engagement, curating relevant inventory. While programmatic is a major buzzword, boutique firms with access to top-tier inventory are at a clear advantage, because with human-powered optimization, you can tell what's relevant to brands and their audiences and help place media in relevant top tier and up-and-coming media mixes, ensuring consistent branding and engagement.

That’s the best approach to the current media environment. Years ago, magazines had long runs in part because there was a lack of competition. TV introduced a faster pace in which new shows could be canceled in a few weeks or even a single night if they didn’t catch on.

Digital media is something different, a dynamic marketplace in which new players are entering all the time while old ones are constantly reinventing themselves. Past performance is no indicator of future results, but keeping a close eye on the market definitely helps to make smart decisions.